Mark Baker wrote:
On 6/14/06, Jeffrey Winter <[EMAIL PROTECTED]> wrote:
Could not the processing model of Atom be clearly specified to
allow such things as atom:id overrides, in the full
spirit of TAG Finding 12, "Authoritative Metadata",
section 3.1 "Role of Internet Media Types"?
http://www.w3.org/2001/tag/doc/mime-respect-20060412
I don't think so. At least, I can't think of way to get around PUT
semantics in that regard.
This is what the Title: header does. It puts that metadatum in a place
where it's authoritative. Stuff buried in an atom:entry is not
Authoritative Metadata per that finding. Otoh, we need servers to handle
items like atom:entry/atom:id in order to have the APP function sanely,
ie, I think some amount of tying the APP into Atom as XML is inevitable.
It's a fine line - if we go too far into digging out protocol data from
the Atom entities we'll make the kinds of mistakes the WS people have by
driving dependencies up from the entity into HTTP (which is precisely
backways). All the syntax WG has to do is add a new mandatory
field/attribute and every APP server/client on the planet will be
broken- supposedly that can't happen, but you get the idea.
cheers
Bill
